Microwave ablation thermal field simulation method and microwave ablation simulation system based on neural network

Abstract

The invention relates to a microwave ablation thermal field simulation method based on a neural network and a microwave ablation system, and the method comprises the steps: collecting clinical data, obtaining tumor information, tumor surrounding environment information, ablation parameters and an actual ablation range of each tumor as a training set, and carrying out the simulation of an ablation thermal field through a neural network model. The neural network model is utilized to select proper ablation parameters including the needle type of the ablation needle, the ablation time, the ablation power and the like through the tumor information and the tumor surrounding environment information, so that an ablation thermal field just envelops a focus, insufficient ablation or excessive ablation can be avoided, and harm to a patient is reduced.

Description

Technical Field

[0001] The invention relates to a microwave ablation thermal field simulation method based on a neural network and a microwave ablation system, belonging to the technical field of microwave ablation. Background Art

[0002] Microwave ablation technology is widely used in the treatment of various solid tumors. When formulating an ablation plan, doctors cannot accurately evaluate ablation parameters, such as ablation time, ablation power, and the best ablation needle. Therefore, in order to avoid incomplete ablation, excessive ablation range, and damage to safe organs, CT images are used to evaluate the ablation status multiple times during surgery. In order to ensure that the ablation results meet expectations, the number of CT scans is generally increased to improve the accuracy of judgment, which will increase the risk of patients being exposed to radiation, increase the economic burden on patients, and increase the complexity of doctors' surgical procedures.

[0003] According to the applicant's understanding, the current microwave ablation system automatically selects ablation parameters based on the medical images of the tumor, and estimates the intraoperative ablation effect through the preoperative simulated ablation thermal field images. However, since the simulated ablation thermal field is obtained under relatively theoretical conditions, it does not take into account the environmental parameters surrounding the tumor (including blood vessel size, tracheal size, the distance between the tumor and the blood vessel, the distance between the tumor and the trachea, the distance between the tumor and the nearest organ, etc.), and there is usually a large difference between the actual ablation effect.

[0004] The applicant searched and found that the Chinese patent document with publication number CN115607269A disclosed a microwave ablation thermal field simulation method and system, which took into account the influence of the heat carried away by blood flow on the ablation results. However, the comparative document 1 still did not take into account the influence of the environmental parameters surrounding the tumor on the ablation electromagnetic field. Therefore, the simulated ablation thermal field still did not match the actual ablation thermal field. The inaccurate simulated ablation thermal field ultimately affected the accurate selection of ablation parameters in the actual operation, and there was a defect of insufficient ablation or excessive ablation affecting other normal tissues, causing unnecessary harm to the patient. Summary of the invention

[0005] The technical problem to be solved by the present invention is to propose a microwave ablation thermal field simulation method and a microwave ablation system for accurately selecting ablation parameters to avoid insufficient ablation or excessive ablation.

[0006] In order to solve the above technical problems, on the one hand, the present invention provides a microwave ablation thermal field simulation method based on a neural network, comprising the following steps: Collecting clinical data, the clinical data including a tumor model, an organ model corresponding to the tumor, a blood vessel model, a trachea model, ablation parameters, and an actual ablation range, wherein the blood vessels and trachea are located within a preset range of the tumor; Preprocessing the clinical data to obtain tumor information, tumor surrounding environment information, ablation parameters and actual ablation range of each tumor as a training set, wherein the tumor information includes the size of the tumor, the type of organ corresponding to the tumor and the size of the organ corresponding to the tumor, and the tumor surrounding environment information includes the size of the blood vessels, the size of the trachea, the distance between the tumor and the blood vessels, and the distance between the tumor and the trachea; Constructing a neural network model for selecting ablation parameters, and training the model using the training set; Reconstruct the patient's lesion into a three-dimensional model using medical images, find the tumor target tissue segmentation model of the lesion and the organ model at the lesion, and obtain the tumor information of the lesion and the surrounding environment information of the tumor; record the range that just envelops the lesion as the theoretical ablation range of the lesion; The tumor information of the lesion, the tumor surrounding environment information and the theoretical ablation range are input into the neural network model to obtain the required ablation parameters and complete the simulation of the microwave ablation thermal field.

[0007] On the other hand, the present invention also provides a microwave ablation simulation system, including a medical image three-dimensional reconstruction module, a lesion and surrounding environment information extraction module, an ablation thermal field recommendation module and an ablation module; The medical image three-dimensional reconstruction module is configured to reconstruct the medical image of the patient's lesion into a three-dimensional model; The lesion and surrounding environment information extraction module is configured to extract a tumor target tissue segmentation model of the lesion and an organ model at the lesion from the three-dimensional model to obtain tumor information of the lesion and tumor surrounding environment information, wherein the tumor information includes the size of the tumor, the organ type corresponding to the tumor, and the organ size corresponding to the tumor, and the tumor surrounding environment information includes the size of the blood vessels, the size of the trachea, the distance between the tumor and the blood vessels, and the distance between the tumor and the trachea; The ablation thermal field recommendation module is configured to recommend an ablation thermal field using the microwave ablation thermal field simulation method; The ablation module is configured to use the ablation thermal field selected by the ablation thermal field recommendation module to plan a surgical plan.

[0008] The beneficial effects brought about by the present invention are: the present invention utilizes a neural network model to select appropriate ablation parameters, including the needle type, ablation time, and ablation power of the ablation needle, through tumor information and tumor surrounding environment information, so that the ablation thermal field just envelopes the lesion, thereby avoiding insufficient or excessive ablation and reducing harm to the patient. DETAILED DESCRIPTION

[0009] Microwave ablation for lung tumors is achieved by releasing microwave magnetic fields to generate temperatures above 60°C in tumor tissues, causing tissue coagulation, dehydration, and necrosis. However, in clinical treatment, microwave ablation lacks appropriate power and time parameters, resulting in poor ablation or excessive ablation. This embodiment uses the ablation treatment of liver tumors as an example to simulate the ablation thermal field based on a neural network, and uses the above ablation parameters to perform microwave ablation on the lesions.

[0010] Embodiment 1: This embodiment shows a method for simulating the thermal field of microwave ablation based on a neural network, comprising the following steps: 1) Collect clinical data, including the tumor model, the organ model corresponding to the tumor, the blood vessel model, the trachea model, the ablation parameters and the actual ablation range, wherein the blood vessels and trachea are located within the preset range of the tumor, and the preset range is usually not less than 2 cm from the edge of the tumor. The ablation parameters include the needle type of the ablation needle, the ablation time and the ablation power.

[0011] 2) Preprocessing the clinical data to obtain the tumor information, tumor surrounding environment information, ablation parameters and actual ablation range of each tumor as a training set, wherein the tumor information includes the size of the tumor, the type of organ corresponding to the tumor and the size of the organ corresponding to the tumor, and the tumor surrounding environment information includes the size of the blood vessels, the size of the trachea, the distance between the tumor and the blood vessels, and the distance between the tumor and the trachea. Preprocessing includes washing abnormal data and duplicate data, etc. Preferably, the tumor surrounding environment information also includes the distance between the tumor and the nearest organ.

[0012] 3) Construct a neural network model for selecting ablation parameters and train it using the training set. The neural network model can be a UNet or nnUNet model, and the training framework can be a TensorFlow, PyTorch, or Keras framework.

[0013] 4) Import the patient's CT / MRI images. Medical images can be obtained through USB flash drives, CDs, or direct connection to the PACS system. The patient's lesions are reconstructed into a three-dimensional model using the medical images to find the tumor target tissue segmentation model of the lesion and the organ model at the lesion, and the tumor information of the lesion and the surrounding environment information of the tumor are obtained; the range that just envelops the lesion is recorded as the theoretical ablation range of the lesion. Reconstructing a three-dimensional model and segmenting the tumor information of the lesion and the surrounding environment information of the tumor are both existing technologies and will not be described in detail. Preferably, when extracting the surrounding environment information of the tumor, the minimum range to be extracted is 2 cm outward from the edge of the tumor.

[0014] When the tumor surrounding environment information includes the distance between the tumor and the nearest organ, the organ closest to the lesion is extracted from the three-dimensional model, and the distance between the two is obtained and applied to the prediction of the neural network model.

[0015] 5) The tumor information of the lesion, the tumor surrounding environment information and the theoretical ablation range are input into the neural network model to obtain the required ablation parameters and complete the simulation of the microwave ablation thermal field.

[0016] This embodiment can also be improved as follows: after surgery, the actual ablation range of the lesion obtained based on clinical data is compared with the theoretical ablation range of the lesion. If the accuracy does not meet the preset requirements, the tumor information, tumor surrounding environment information, actual ablation range and ablation parameters of the lesion are used as training samples to further train the neural network model. By continuously increasing the number of training samples, the accuracy of the neural network model is continuously improved.

[0017] This embodiment provides a high-precision simulated thermal field for ablation surgery, and as the number of cases continues to increase, the simulation accuracy will become higher and higher, which can provide a reliable reference basis for ablation surgery, improve the safety and effectiveness of the surgery, and has important clinical value.

[0018] Embodiment 2: This embodiment shows a microwave ablation simulation system, including a medical image three-dimensional reconstruction module, a lesion and surrounding environment information extraction module, an ablation thermal field recommendation module and an ablation module.

[0019] The medical image three-dimensional reconstruction module is configured to reconstruct the medical image of the patient's lesion into a three-dimensional model; The lesion and surrounding environment information extraction module is configured to extract a tumor target tissue segmentation model of the lesion and an organ model at the lesion from the three-dimensional model to obtain tumor information of the lesion and tumor surrounding environment information, wherein the tumor information includes the size of the tumor, the organ type corresponding to the tumor, and the organ size corresponding to the tumor, and the tumor surrounding environment information includes the size of the blood vessels, the size of the trachea, the distance between the tumor and the blood vessels, and the distance between the tumor and the trachea; The ablation thermal field recommendation module is configured to recommend an ablation thermal field using the microwave ablation thermal field simulation method described in Example 1; The ablation module is configured to use the ablation thermal field selected by the ablation thermal field recommendation module to plan the surgical plan. The surgical plan planning in this embodiment can be to select a suitable ablation needle type, puncture channel, ablation time and power, etc., to reduce surgical risks.

[0020] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A microwave ablation thermal field simulation method based on neural network, comprising the following steps: Collecting clinical data, the clinical data including a tumor model, an organ model corresponding to the tumor, a blood vessel model, a trachea model, ablation parameters, and an actual ablation range, wherein the blood vessels and trachea are located within a preset range of the tumor; Preprocessing the clinical data to obtain tumor information, tumor surrounding environment information, ablation parameters and actual ablation range of each tumor as a training set, wherein the tumor information includes the size of the tumor, the type of organ corresponding to the tumor and the size of the organ corresponding to the tumor, and the tumor surrounding environment information includes the size of the blood vessels, the size of the trachea, the distance between the tumor and the blood vessels, and the distance between the tumor and the trachea; Constructing a neural network model for selecting ablation parameters, and training the model using the training set; Reconstruct the patient's lesion into a three-dimensional model using medical images, find the tumor target tissue segmentation model of the lesion and the organ model at the lesion, and obtain the tumor information of the lesion and the surrounding environment information of the tumor; record the range that just envelops the lesion as the theoretical ablation range of the lesion; The tumor information of the lesion, the tumor surrounding environment information and the theoretical ablation range are input into the neural network model to obtain the required ablation parameters and complete the simulation of the microwave ablation thermal field.

2. The method for simulating thermal field of microwave ablation based on neural network according to claim 1, characterized in that: The tumor surrounding environment information also includes the distance between the tumor and the nearest organ.

3. The method for simulating thermal field of microwave ablation based on neural network according to claim 1, characterized in that: When extracting the tumor surrounding environment information, the minimum extraction range is 2 cm outside the tumor edge.

4. The method for simulating thermal field of microwave ablation based on neural network according to claim 1, characterized in that: The neural network model is a UNet or nnUNet model, and its training framework is a TensorFlow, PyTorch or Keras framework.

5. The method for simulating thermal field of microwave ablation based on neural network according to claim 1, characterized in that: After the operation, the actual ablation range of the lesion obtained based on the clinical data is compared with the theoretical ablation range of the lesion. If the accuracy does not meet the preset requirements, the tumor information, tumor surrounding environment information, actual ablation range and ablation parameters of the lesion are used as training samples to further train the neural network model.

6. A microwave ablation simulation system based on a neural network, characterized in that: It includes a medical image 3D reconstruction module, a lesion and surrounding environment information extraction module, an ablation thermal field recommendation module, and an ablation module; The medical image three-dimensional reconstruction module is configured to reconstruct the medical image of the patient's lesion into a three-dimensional model; The lesion and surrounding environment information extraction module is configured to extract a tumor target tissue segmentation model of the lesion and an organ model at the lesion from the three-dimensional model to obtain tumor information of the lesion and tumor surrounding environment information, wherein the tumor information includes the size of the tumor, the organ type corresponding to the tumor, and the organ size corresponding to the tumor, and the tumor surrounding environment information includes the size of the blood vessels, the size of the trachea, the distance between the tumor and the blood vessels, and the distance between the tumor and the trachea; The ablation thermal field recommendation module is configured to recommend an ablation thermal field using the microwave ablation thermal field simulation method according to any one of claims 1 to 6; The ablation module is configured to use the ablation thermal field selected by the ablation thermal field recommendation module to plan a surgical plan.